Orrester predicts that the low-code market will grow to $30 billion by 2028. According to the report by Mordor Intelligence, the low-code development platform market size is expected to be USD 16.17 billion by 2024 and expected to reach USD 62.15 billion by 2029, registering a CAGR of 30.90% during the forecast period (2024-2029).
In order to solve a technical problem, Devin borrowed the account of its creator, communicated with the CTO of the customer company, and adjusted the code scheme according to the reply.
While everyone is still amazed, iFLYTEK's AI programmers are not only used in-house, but also have been "employed" in more than 100 companies such as JD.com, iSoftStone, and Bank of Communications, which can not only understand complex programming logic, but also automatically generate high-quality code, helping developers improve code development efficiency and accelerate the construction of the digital world.
[The strongest code "assisted" in more than 100 enterprises "onboarding"]
Code capability is an important indicator of the "smartness" of large models, and has also become the focus of general large models at home and abroad, such as OpenAI GPT, Google Gemini, iFLYTEK Xinghuo, and Wenxin Yiyan.
Since its release last year, iFLYTEK Xinghuo has focused on upgrading code capabilities as one of the seven core capabilities, and systematically released a series of code functions for code generation, completion, error correction, interpretation and unit test generation in August last year. According to the news released by iFLYTEK on January 30 this year, the code capability of iFLYTEK Xinghuo in the latest V3.5 version has reached 96% of GPT-4 Turbo, and it has surpassed Python, the most commonly used programming language in scientific research.
A front-end and back-end development engineer of an AGV company said that since using the Xinghuo code model, his work efficiency has been greatly improved, and the biggest benefit is that he only needs to describe the desired page layout effect clearly, and the code can be automatically generated, saving a lot of time to type code. He estimates that development efficiency has increased by about 30%-40%.
He also pointed out that the Xinghuo code model is more friendly to Chinese programmers, in addition to a stronger understanding of Chinese, the way of use is also more in line with the thinking habits of Chinese people, it can report errors in Chinese, and also reduces the debugging cost of developers.
A software engineer with 20 years of code said that he has used a lot of AI tools, and now mainly uses the Xinghuo code model and Copilot, he believes that the two are basically an echelon in semantic understanding, and they are compatible with various development platforms in the form of plug-ins, which is very convenient to call. It is worth mentioning that the Xinghuo code model fully opens the AI dialogue assistant window to provide real-time assistance, which greatly simplifies the process of code finding and problem solving.
In a leading state-owned bank, the Spark code model has been "put on the job" and has become the code "partner" of more than 3,000 software developers in the bank, achieving a 40% code adoption rate and a 15% improvement in R&D efficiency, effectively improving development efficiency.
At Beijing University of Posts and Telecommunications, based on the code capabilities of iFLYTEK Xinghuo, BUPT has developed an intelligent programming teaching application platform "Code", which provides real-time, personalized and heuristic programming tutoring services for the needs and pain points of students in the process of programming teaching, and at the same time flexibly and efficiently supports the teaching work of teachers, which has been used by teachers and students of the whole school.
In the mainland, there are about 8 million software employees, and the revenue of the software industry in the mainland will reach 10.8 trillion yuan in 2022, accounting for 8.93% of GDP, and it is predicted that the revenue of the software industry will double to 20 trillion yuan by 2028. In such a trillion-level market, the rapid changes in market demand and frequent changes in customer needs bring challenges to development efficiency, code quality problems, and pain points such as "code debt" caused by the maintenance of old code, all of which mean that AI programmers have more room for development, and now is just the beginning.
2024 will be a key year for the application of large models, and the code field has also become the focus of industry competition. It is worth noting that iFLYTEK recently announced that it will release the iFLYTEK Xinghuo V4.0 benchmarking GPT-4 in June this year, coupled with the first domestic large-scale domestic large-scale computing platform built by it with Huawei last year, both computing power and model parameters will be further upgraded. In the face of international competitors such as Copilot, can the Spark code model successfully break through? Let's wait and see!
disclaimer
The content of this article related to listed companies is the author's personal analysis and judgment based on the information publicly disclosed by listed companies in accordance with their legal obligations (including but not limited to temporary announcements, periodic reports and official interactive platforms, etc.), and the information or opinions in this article do not constitute any investment or other business advice, and Market Value Watch does not assume any responsibility for any actions arising from the adoption of this article.
——END——